Modeling the Clonal Evolution of Cancer from Next Generation Sequencing Data

Wei Jiao, Shankar Vembu, Amit G. Deshwar, Lincoln Stein, Quaid Morris
(Submitted on 11 Oct 2012)

We consider the problem of inferring the clonal evolutionary structure of cancer cells from high-throughput next generation sequencing data. We address this problem using statistical machine learning to infer a relational clustering of objects, where the clusters are connected in the form of a rooted tree. We present a hierarchical Bayesian mixture model that uses a non-parametric prior over trees to automatically estimate the number of clones (clusters) and their clonal frequencies (cluster means) in the population, and to identify the phylogenetic relationship between these subclones. Experiments on three real data sets comprising 12 tumor samples from triple-negative breast cancer, acute myeloid leukemia and chronic lymphocytic leukemia patients demonstrate the efficacy of our method.